India's artificial intelligence (AI) ambitions have triggered a surge in demand for software developers, data scientists and machine learning engineers.

But as companies increasingly deploy AI across banking, healthcare, customer service and other sectors, another talent shortage may be quietly emerging—professionals who can test whether these systems are accurate, secure and reliable before they reach users.

Harry Rao, Founder and CEO of TestGrid, believes the conversation around India's AI workforce has largely focused on building AI systems, while overlooking the equally important task of evaluating them.

"When you hear about India's AI talent needs, the conversation often centres on software developers, data scientists and Machine Learning (ML) engineers. That focus is understandable," Rao said.

He pointed to NASSCOM's 2023 State of Data Science \& AI Skills in India report, which projected that demand for AI and data science professionals in India's technology sector would rise from around 629,000 in 2022 to nearly 1.13 million by 2026.

"But building AI systems is only one part of the workforce challenge," Rao said.

"A second requirement is becoming harder to ignore: organisations need professionals who can determine whether those systems are accurate, secure, fair and dependable enough for real-world use."

According to Rao, enterprise adoption of AI is moving faster than companies' ability to evaluate it, creating what could become India's next technology talent gap.

"From what we see across enterprise quality teams, AI adoption is advancing faster than many companies' ability to evaluate it. India may, therefore, be on the verge of facing another talent gap—this time, in AI quality engineering," he said.

WHY TESTING AI IS DIFFERENT FROM TESTING SOFTWARE

Traditional software testing follows a relatively predictable process. A user performs an action, the software returns an expected outcome and testers verify whether it meets predefined requirements.

AI systems, however, behave differently.

The same prompt can generate different responses, and while those answers may appear convincing, they can still contain factual inaccuracies or unsupported claims. Their performance may also change over time as models are updated, source data evolves or user behaviour shifts.

That means quality teams can no longer rely solely on conventional testing or test automation practices.

Instead, they need to evaluate whether training and evaluation data are representative, current and relevant, whether AI models treat similar users consistently, whether attackers can manipulate prompts to bypass safeguards or expose sensitive information, when decisions should be escalated to a human, and how organisations will detect performance declines after deployment.

"A final round of interface testing won't answer all these questions," Rao said.

"Quality engineering must cover the AI application as a whole, including data, model behaviour, interfaces, safeguards and post-release monitoring."

INDIA'S AI CHALLENGE GOES BEYOND THE MODEL

India presents unique challenges for AI quality assurance because of its linguistic diversity, regional differences and varying levels of digital infrastructure.

According to Rao, organisations need test cases that reflect how people actually use AI systems rather than ideal scenarios.

For example, a financial assistant should understand requests that combine Hindi and English, distinguish a balance enquiry from an unauthorised transaction complaint, and accurately interpret everyday language, even when users misspell words or describe problems informally.

Healthcare applications require even greater scrutiny.

An AI-powered medical summarisation system may generate a polished report while unintentionally omitting an allergy, a dosage change or another clinically important observation. Evaluating such systems requires medical expertise, clearly defined safety standards and processes that allow uncertain cases to be escalated to qualified professionals.

Voice-based AI systems also need to be tested across regional accents, speech patterns, background noise and different audio conditions. Consumer-facing applications must continue to perform reliably on affordable smartphones, older operating systems and unstable mobile networks.

WHY TEST AUTOMATION ALONE IS NO LONGER ENOUGH

While test automation remains a critical part of software quality engineering, Rao said AI applications demand a much broader set of skills.

Traditional quality engineering capabilities such as software reliability, performance testing, release quality and test automation will continue to be important. However, AI quality engineering now requires professionals to evaluate model accuracy, identify unsupported claims, assess bias across different populations, test adversarial prompts, uncover security vulnerabilities, define human escalation rules and monitor system performance after deployment.

According to Rao, this expands quality engineering beyond software testing into data literacy, cybersecurity awareness, AI model evaluation and domain expertise.

"AI quality engineering combines software testing with data literacy, security awareness, model evaluation and domain judgement—a mix of capabilities that organisations may struggle to find," he said.

THE NEXT BIG AI JOB MAY NOT BE CODING

Rao believes companies should start building these capabilities before AI systems begin failing in production.

"That's exactly why you shouldn't wait for AI systems to fail in production before building these capabilities," he said.

He added that quality engineers should become involved much earlier in the AI development lifecycle, helping define what systems should do, where they could cause harm and how acceptable performance will be measured.

Simply hiring a traditional software tester and adding "AI" to the job description, he argued, will not be enough.

Organisations need to decide which capabilities can be developed internally and where specialised expertise in data, security or AI model evaluation will be required. At the same time, universities and training institutions should begin treating AI evaluation as a dedicated discipline rather than an extension of conventional software testing.

"India will continue to need people who can build AI systems. But its ability to deploy them responsibly will also depend on professionals who can test their behaviour, identify their limits and decide how much trust they deserve," Rao said.

- Ends

Published On:

Jul 21, 2026 17:14 IST